Computational Imaging using Virtualized Pattern Projection

Numerous imaging applications rely on spatially patterning the incident irradiance within the imager field of regard. A common approach to engineering sinusoidal illumination is the interference of two mutually coherent beams. The concept can be generalized to adapt surfaces such as walls/floors to behave as sources of patterned illumination. In one embodiment, the surface of convenience may be illuminated by two tightly focused spots that behave as point-sources in a Young’s double pinhole experiment. A detector/object positioned beyond the surface observes a sinusoidal pattern, with small deviations arising from the optical roughness of the underlying surface. Current techniques for processing images acquired under sinusoidal illumination may be adapted to accommodate such deviations. The above notion of Virtualized Pattern Projection has numerous applications including indirect illumination of objects hidden from view, ranging using Structured Light and Optical Super-Resolution.